60 views (last 30 days)

Show older comments

Hi, I am really frustrated about that issue, because it seems to me, that it is really easy, but I dont get it running...

So I try to write some cells in an Excel sheet, in a specific range. The range is supplied to the xlswrite or writetable function in the Excel style, i.e. A1:AB15. I would like to create a string that specifies the range dynamically. So I need a function, that converts numbering (1,2,3 ...) to "excel alphabetical numbering" (A,B,C, ..., Z, AA, AB...). I tried a lot around with some base 26 and base 27 stuff. But it still doesn't give me satisfying results. In the end, I don't really care anymore about the Excel table numbering stuff, i just want to get this function alphabetStr = num2ExcelAlpha(num) to produce the correct string.

My initial thoughts were: The alphabet is some kind of base 26 enumeration (sorry for the wrong terms but I am an engineering grad, not a maths grad...). So basically I can get the number of digits of my "alphabetical number" with some log to the base 26 stuff. It didn't work for me.

In the end I am counting the digits kind of "manually". I bet there is something easier with logarithms, but I didn't figure it out.

digits = 0;

while num > 0

num = num - 26^(digits+1);

digits = digits+1;

end

Now, that we have the number of digits, I would like to assign every number a string consisting of digits characters:

alph = 'A':'Z'

str = [];

for i = 1:digits

idx = floor(num/26^(digits-i));

if idx == 0

idx = 26;

end

str = [str alph(idx)];

num = num - floor(num/26^(digits-i)); %probalby *26^(digits-i); probably ceil(); something is wrong here

if num == 0

num = 26^(digits-i-1);

end

end

You see, this is already really confusing what I did there, but this is already my 50th attempt, and it still doesn't give me the right characters consistantly. Mostly, whether the strings with 'A' or 'Z' are missing, wrong, or the index is 0 and it throws an error.

- Is there an easier solution to this?
- If not included in 1: What did I miss regarding the maths part of this problem?

Thanks a lot for your help!

dpb
on 19 Jan 2021

Edited: dpb
on 19 Jan 2021

function rnge=xlsAddr(row,col)

% Build Excel cell address from row, column

%

% RNGE=XLSADDR(ROW,COL) will return an Excel cell address

% formed from the input ROW,COL values. Either input may be

% string or numeric and will be converted to canonical form

if isnumeric(col)

if ~isscalar(col), error('Input Column Not Scalar'), end

N=26+(col==26); % kludge to fix up rollover Z-->AA

rnge=num2str('A'+[fix(col/N) rem(col,N)]-1,'$%c%c');

rnge(rnge=='@')=[]; % cleanup for single character

else

rnge=['$' col];

end

if isnumeric(row)

if ~isscalar(row), error('Input Row Not Scalar'), end

rnge=[rnge num2str(row,'$%d')];

else

row=num2str(row,'$%d');

if ~all(ismember(row,'0':'9')), error('Invalid Excel Address: Row not numeric'), end

rnge=[rnge row];

end

is my utility function I threw together...it seems to work ok athough I'm sure it could be made much more elegant.

As you can see, I also had some issues with the wraparound that I just kludged...

Stephen Cobeldick
on 20 Jan 2021

"it seems like the 0 is missing. But not completely missing, since you could still imagine it as "initial" digit, i.e. A = 0A = 1 = 01."

No, the zero digit really is missing entirely. How does adding an implicit leading zero make any difference to the skipped index for 27 (i.e. "A0")? Excel jumps straight from Z to AA (and from AZ to BA, etc.). Every zero digit is missing.

Adding infinite implicit leading zeros makes zero(!) difference, is just an irrelevant distraction from the actual problem.

"I wasn't aware of that, either, not really being an Excel person."

I have often processed reasonably large data files, for which knowing the actual limits of Excel became quite important (and ergo, getting the indexing right). These are not just hypothetical concerns either:

Fangjun Jiang
on 19 Jan 2021

- In old versions of MATLAB, you can find that function inside xlswrite.m. I used to copy that for my own use.
- In R2019b, it seems it is using "import matlab.io.spreadsheet.internal.columnLetter". You could do the same.
- dec2base() is the related function. You want base to be 'ABCDEFGHIJKLMNOPQRSTUVWXYZ' which is 26. But when you specify the base to be 26, the function output string of '0123456789ABCDEFGHIJKLMNOP'. You could do some further conversion.

Stephen Cobeldick
on 19 Jan 2021

The missing zeros make it more complex than just converting the base.

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!